center and truncate should be grapheme cluster and unicode width aware

[Kijewski]

Let's have a look at the letter 'ễ' in the surname "Nguyễn". You can either find it as a single composed unicode character U+1EC5. Or decomposed as "e\u{302}\u{303}". When truncating a text, the letter ễ should stay ễ, and not be truncated to ê or e. Or take the emoji "👯‍♂️", which is composed of "\u{1f46f}\u{200d}\u{2642}\u{fe0f}", a sequence that must not be split up.

It would be nice if one could make |center and |truncate understand unicode widths (ễ = 1 display character; 👯‍♂️ = 2 display characters), and grapheme clusters. It should be opt in, because the lookup tables are big.

Maybe, instead of modifying the existing functions, new ones should be introduced.